AI Regulation Tracker  /  Draft consultation

India's Supreme Court Publishes Draft Rules for AI Use in Courts, 2026

On 3 June 2026, the Supreme Court of India's AI Committee published a preliminary draft rulebook for how artificial intelligence may be used across the country's courts. It is a consultation draft, not law, and comments are due by 15 July 2026.

By Anthony Guerriero  ·  The Leveraged Years Regulatory Research  ·  First published July 14, 2026  ·  Last updated July 14, 2026

What the draft actually does

Start with what this is. On 3 June 2026 the Supreme Court published a notice that reads: "a preliminary draft of 'Regulations for Use of Artificial Intelligence (AI) in Courts, 2026' has been prepared under the aegis of the Artificial Intelligence Committee, Supreme Court of India." The notice says the aim is to govern AI in courts, "grounded in the principles of human primacy, transparency, accountability, data protection, and judicial independence."

The document is a draft. Its own commencement clause makes that plain. Regulation 1 says the rules "shall come into force" for the Supreme Court "on such date as the Chief Justice of India may, by notification in the Official Gazette, appoint," and for each High Court on a date the respective Chief Justice appoints. Different provisions can start on different dates. Until those notifications happen, none of this operates.

The consultation window matters here. The first notice asked for comments by 20 June 2026. A second notice, dated 16 June 2026, extended it. That notice states the "deadline for the same stands extended till 15.07.2026 as per directions of the Artificial Intelligence (AI) Committee, Supreme Court of India." Comments go to the Member Secretary at office.regcc@sci.nic.in.

Now the substance. The draft splits AI use into permitted and prohibited categories, and the prohibited list is where the hard lines are.

Regulation 20 lists uses that are "strictly prohibited in all Court processes" and calls them "absolute and non-derogable." Three of them are worth reading closely if your work touches Indian litigation:

• No decision by machine. Under 20(1)(b), "no judicial outcome (including any judgment, order, or finding of fact or law) shall be reached through Algorithmic Decision-Making alone or solely on the basis of AI-generated information," and "the human judicial authority shall be the determinative authority in all adjudicative decisions." Regulation 20(1)(c) adds that any AI output on an adjudicative or sentencing question "shall be treated as advisory only."
• No risk scoring. Under 20(1)(d), "no AI System shall be used for Risk Scoring for any purpose in Court processes, including the assessment of flight risk, prediction of recidivism, evaluation of bail eligibility, or determination of the credibility of parties or witnesses." The draft defines Risk Scoring as assigning a score that "purports to estimate the probability of that person engaging in a specified future behaviour."
• No AI evidence by stealth. Under 20(1)(h), "no AI-generated output shall be submitted to a Court as an independent source of evidence without full and transparent disclosure of its AI-generated character."

The permitted side, Regulation 19, is broad but leashed. It allows AI for case management, cause-list preparation, transcription "subject to mandatory review and certification of accuracy," translation "subject to human verification," and "legal research, precedent retrieval, citation verification and document summarisation." Every one of those sits under "prior approval in writing by the Appropriate Authority" and human oversight. The pattern is consistent. AI drafts and retrieves; a person checks and decides.

Who is affected, and how

Litigators and advocates. The disclosure duty is the piece that reaches your daily filings. Regulation 43(3) says that where an AI tool is used "by any party or his legal representative in the preparation or submission of any document, pleading, or evidence, the AI-assisted character of such material shall be disclosed to the Court at the time of submission" through a declaration in a prescribed format (Annexure I). Then comes the accountability clause. Regulation 43(6) provides that if a submission "is found to be fabricated, false, misleading, or inaccurate by reason of its AI-generated character, the person submitting the same shall bear full responsibility therefor and shall not be entitled to rely upon the character of the AI output as a defence."

Read that last part twice. "The AI made it up" would not be a defence. The person who filed it owns it.

In-house counsel. If your company litigates in India, the exposure is indirect but real. If adopted substantially as drafted, your outside counsel would carry the filing-disclosure and verification burden, and your matters would run inside courts operating under this framework once notified. Worth a line in your outside-counsel guidelines about how AI-assisted work product gets verified before filing.

Legal-technology vendors. Chapter VI is aimed at you. Regulation 46(1) says no vendor may provide any AI service to court processes "without the prior written approval of the Appropriate Authority." Regulation 46(4) then requires that every vendor agreement include a specific set of terms: "obligations of disclosure, incident reporting, and cooperation with audits"; "the right of the AI Secretariat to audit and inspect the relevant AI System and its underlying data"; "mandatory indemnity clauses protecting the Court from liability for harms caused by defects in vendor-supplied AI Systems"; and "clear contractual allocation of liability between the Court and the vendor in the event of AI-related incidents, data breaches, or harm to litigants or third parties." Regulation 46(6) requires reporting any breach or AI incident "without delay." If you sell to courts, your contract template and your audit posture are the work.

Judges and court staff. The draft keeps the judge in the determinative seat and layers oversight around the tools, including an AI Register, periodic audits, an AI Incident Database, and an annual transparency report per court.

So what for your practice

Here is the honest read. Nothing on this page requires you to do anything today, because the draft is not in force. But two habits are worth building now, because they cost little and they line up with where this is going.

First, keep a verification trail on anything AI touched in a filing. Citation checks, source confirmation, a note of who reviewed it. If a disclosure-and-responsibility regime like Regulation 43 lands, that trail is your protection. It is also just good practice regardless of what any court adopts.

Second, if you are on the vendor side, read Regulation 46 against your current contracts. Indemnity for AI defects, audit access, and incident reporting are the three terms most likely to be missing. Those are slow to negotiate, so starting early is cheap insurance.

The framing throughout the draft is augmentation, not replacement. The machine assists, retrieves, transcribes, and translates. A human verifies and decides. That is a workable way to think about AI in any regulated practice, not only litigation in India.

The cross-border hook

For global firms, the value of this document is as a template signal. India is a large common-law jurisdiction writing down, in one place, a set of prohibitions that other courts have so far handled case by case. The specific bans on AI-only outcomes and on risk scoring for bail, recidivism, and credibility are the kind of provisions that travel. If you advise clients across multiple legal systems, this draft is a useful map of the lines regulators and courts are starting to draw. Watch what survives consultation, because the surviving version is the one worth benchmarking against.

One caution for anyone briefing a client. This draft is separate from the Indian Supreme Court's earlier ruling that citing AI-generated fake precedents can amount to advocate misconduct and can render a judgment void. That was a decision in an actual case, applied to conduct that already happened. This is a proposed rulebook for future AI use, still out for comment. Do not blur the two. See our related briefing on the misconduct judgment, linked below.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel in the relevant jurisdiction.